Can the NVIDIA H100 SXM 80GB run WeMM-Video-35BA3B?
WeMM-Video-35BA3B·36.2B (5B active)·NVIDIA H100 SXM 80GB·80 GB VRAM
Yes. The NVIDIA H100 SXM 80GB runs WeMM-Video-35BA3B fully in VRAM at FP16 / BF16 (about 72 GB of 80 GB), at roughly 185 tokens/s.
- best quant
- FP16 / BF16
- needs
- 72 GB
- usable
- 80 GB
- est. speed
- ~185 tok/s
Estimates for an 8K context; offload assumes 32 GB of system RAM.Try other contexts on the WeMM-Video-35BA3B page →
Every quantisation of WeMM-Video-35BA3B on the NVIDIA H100 SXM 80GB
| Quant | Memory | Fit | Speed |
|---|---|---|---|
| Q2_K | 13 GB | fits | ~1127 tok/s |
| Q3_K_M | 18 GB | fits | ~758 tok/s |
| Q4_K_M | 23 GB | fits | ~611 tok/s |
| Q5_K_M | 26 GB | fits | ~521 tok/s |
| Q6_K | 30 GB | fits | ~450 tok/s |
| Q8_0 | 39 GB | fits | ~349 tok/s |
| FP16 / BF16 pick | 72 GB | fits | ~185 tok/s |
Speed is a bandwidth-based estimate for generating text (3350 GB/s); prompt processing and long contexts are slower.
Measured speed on the NVIDIA H100 SXM 80GB
No community reports for this pair yet.
Report your own speed →Vision models that fit the NVIDIA H100 SXM 80GB comfortably
All that fit →| Model | Params | Best quant | Memory | Speed |
|---|---|---|---|---|
| 36BA3B | Q8_0 | ≈38 GBFits in memory | ~578 tok/s | |
| 36BA3B | Q8_0 | ≈37 GBFits in memory | ~578 tok/s | |
| 31.3B | Q8_0 | ≈40 GBFits in memory | ~55 tok/s | |
| 27.8B | Q8_0 | ≈31 GBFits in memory | ~62 tok/s | |
| 35.2BA3B | Q8_0 | ≈37 GBFits in memory | ~578 tok/s | |
| 29.8B | Q8_0 | ≈30 GBFits in memory | ~58 tok/s |
Other GPUs that run WeMM-Video-35BA3B
Frequently asked questions
- Can the NVIDIA H100 SXM 80GB run WeMM-Video-35BA3B?
- Yes. The NVIDIA H100 SXM 80GB runs WeMM-Video-35BA3B fully in VRAM at FP16 / BF16 (about 72 GB of 80 GB), at roughly 185 tokens/s.
- How fast does WeMM-Video-35BA3B run on the NVIDIA H100 SXM 80GB?
- An estimate from memory bandwidth gives about 185 tokens/s at FP16 / BF16.
- Which quantisation of WeMM-Video-35BA3B should I use on the NVIDIA H100 SXM 80GB?
- FP16 / BF16: the largest one that fits. Expected quality: lossless compared with the full-precision weights.
Memory and speed are estimates from the model's size and the card's memory bandwidth; real numbers depend on the runtime, drivers and context length. Community figures are medians of user reports.